Malignant astrocytomas are brain tumors that are locally infiltrative and incurable, with poor prognosis for the patient. Despite profound therapeutic implications, the identity of the cell(s) of origin of these tumors has not been rigorously determined. In addition, glioblastoma multiforme, the most prevalent and deadly form of brain tumor, can either progress from lower grade astrocytic gliomas or arise de novo, the mechanisms by which are not well understood. We previously reported mouse models based on conditional inactivation of human astrocytoma-relevant tumor suppressors p53, Nf1, and Pten, wherein through somatic loss of heterozygosity, mutant mice develop tumors that histologically and molecularly resemble human astrocytomas with 100% penetrance. In the present application, we propose experiments designed to investigate the cell(s) of tumor origin and to extend our analyses of tumorigenesis in these mouse models. To this end, Specific Aim 1 rigorously tests our hypothesis that gliomas originate in stem/progenitor cells. Our research design utilizes genetic and stereotactic methods to mutate the tumor suppressor genes specifically in the stem cell niche. We will also use a genetic method to ablate these neural stem cells in our tumor models and then analyze the effect on tumor formation. Specific Aim 2 will exploit our ability to culture fresh tumor tissue as self-renewable neurospheres to further characterize our Nf1;p53;Pten de novo glioma mouse model. We will evaluate their growth and differentiation properties, as well as their tumorigenic potential via transplantation techniques. Additionally, using a stem cell-specific GFP transgenic mouse, we will analyze the expression of candidate cell surface markers with the aim of identifying signature markers that will allow us to prospectively isolate the cancer stem cells. Specific Aim 3 will employ microarray analyses to identify the gene expression profiles that correlate with de novo vs. progressive glioma, using neurospheres derived from Nf1;p53;Pten and Nf1;p53 tumors, respectively. We also propose to analyze tissue from pre-symptomatic mice in order to gain insight into the early molecular events of tumor initiation. Potentially interesting genes will be functionally pursued using RNAi and overexpression techniques. As microRNAs have recently been implicated in glioma, we will also use microarray analysis to identify microRNAs that are differentially regulated in our tumor models and also investigate the role of candidate microRNAs in glioma. Our fully penetrant glioma mouse models are clinically relevant and powerful tools for identifying and functionally characterizing novel genes and pathways that may be therapeutically tractable in human glioma.